| Message ID | 20230103084755.1256479-1-yukuai1@huaweicloud.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | block, bfq: switch 'bfqg->ref' to use atomic refcount apis | expand |
On Tue 03-01-23 16:47:55, Yu Kuai wrote: > From: Yu Kuai <yukuai3@huawei.com> > > The updating of 'bfqg->ref' should be protected by 'bfqd->lock', however, > during code review, we found that bfq_pd_free() update 'bfqg->ref' > without holding the lock, which is problematic: > > 1) bfq_pd_free() triggered by removing cgroup is called asynchronously; > 2) bfqq will grab bfqg reference, and exit bfqq will drop the reference, > which can concurrent with 1). > > Unfortunately, 'bfqd->lock' can't be held here because 'bfqd' might already > be freed in bfq_pd_free(). Fix the problem by using atomic refcount apis. > > Signed-off-by: Yu Kuai <yukuai3@huawei.com> The patch looks good to me. Feel free to add: Reviewed-by: Jan Kara <jack@suse.cz> Honza > --- > Changes from RFC: > - refcount_dec_and_test(bfqg->ref) -> refcount_dec_and_test(&bfqg->ref) > > block/bfq-cgroup.c | 8 +++----- > block/bfq-iosched.h | 2 +- > 2 files changed, 4 insertions(+), 6 deletions(-) > > diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c > index 1b2829e99dad..7d9b15f0dbd5 100644 > --- a/block/bfq-cgroup.c > +++ b/block/bfq-cgroup.c > @@ -316,14 +316,12 @@ struct bfq_group *bfqq_group(struct bfq_queue *bfqq) > > static void bfqg_get(struct bfq_group *bfqg) > { > - bfqg->ref++; > + refcount_inc(&bfqg->ref); > } > > static void bfqg_put(struct bfq_group *bfqg) > { > - bfqg->ref--; > - > - if (bfqg->ref == 0) > + if (refcount_dec_and_test(&bfqg->ref)) > kfree(bfqg); > } > > @@ -530,7 +528,7 @@ static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, struct request_queue *q, > } > > /* see comments in bfq_bic_update_cgroup for why refcounting */ > - bfqg_get(bfqg); > + refcount_set(&bfqg->ref, 1); > return &bfqg->pd; > } > > diff --git a/block/bfq-iosched.h b/block/bfq-iosched.h > index 41aa151ccc22..466e4865ace6 100644 > --- a/block/bfq-iosched.h > +++ b/block/bfq-iosched.h > @@ -928,7 +928,7 @@ struct bfq_group { > char blkg_path[128]; > > /* reference counter (see comments in bfq_bic_update_cgroup) */ > - int ref; > + refcount_t ref; > /* Is bfq_group still online? */ > bool online; > > -- > 2.31.1 >
Hi, Jens 在 2023/01/03 16:47, Yu Kuai 写道: > From: Yu Kuai <yukuai3@huawei.com> > > The updating of 'bfqg->ref' should be protected by 'bfqd->lock', however, > during code review, we found that bfq_pd_free() update 'bfqg->ref' > without holding the lock, which is problematic: > > 1) bfq_pd_free() triggered by removing cgroup is called asynchronously; > 2) bfqq will grab bfqg reference, and exit bfqq will drop the reference, > which can concurrent with 1). > > Unfortunately, 'bfqd->lock' can't be held here because 'bfqd' might already > be freed in bfq_pd_free(). Fix the problem by using atomic refcount apis. > > Signed-off-by: Yu Kuai <yukuai3@huawei.com> Can you apply this patch? Thanks, Kuai > --- > Changes from RFC: > - refcount_dec_and_test(bfqg->ref) -> refcount_dec_and_test(&bfqg->ref) > > block/bfq-cgroup.c | 8 +++----- > block/bfq-iosched.h | 2 +- > 2 files changed, 4 insertions(+), 6 deletions(-) > > diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c > index 1b2829e99dad..7d9b15f0dbd5 100644 > --- a/block/bfq-cgroup.c > +++ b/block/bfq-cgroup.c > @@ -316,14 +316,12 @@ struct bfq_group *bfqq_group(struct bfq_queue *bfqq) > > static void bfqg_get(struct bfq_group *bfqg) > { > - bfqg->ref++; > + refcount_inc(&bfqg->ref); > } > > static void bfqg_put(struct bfq_group *bfqg) > { > - bfqg->ref--; > - > - if (bfqg->ref == 0) > + if (refcount_dec_and_test(&bfqg->ref)) > kfree(bfqg); > } > > @@ -530,7 +528,7 @@ static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, struct request_queue *q, > } > > /* see comments in bfq_bic_update_cgroup for why refcounting */ > - bfqg_get(bfqg); > + refcount_set(&bfqg->ref, 1); > return &bfqg->pd; > } > > diff --git a/block/bfq-iosched.h b/block/bfq-iosched.h > index 41aa151ccc22..466e4865ace6 100644 > --- a/block/bfq-iosched.h > +++ b/block/bfq-iosched.h > @@ -928,7 +928,7 @@ struct bfq_group { > char blkg_path[128]; > > /* reference counter (see comments in bfq_bic_update_cgroup) */ > - int ref; > + refcount_t ref; > /* Is bfq_group still online? */ > bool online; > >
On Tue, 03 Jan 2023 16:47:55 +0800, Yu Kuai wrote: > The updating of 'bfqg->ref' should be protected by 'bfqd->lock', however, > during code review, we found that bfq_pd_free() update 'bfqg->ref' > without holding the lock, which is problematic: > > 1) bfq_pd_free() triggered by removing cgroup is called asynchronously; > 2) bfqq will grab bfqg reference, and exit bfqq will drop the reference, > which can concurrent with 1). > > [...] Applied, thanks! [1/1] block, bfq: switch 'bfqg->ref' to use atomic refcount apis commit: 216f764716f34fe68cedc7296ae2043a7727e640 Best regards,
diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c index 1b2829e99dad..7d9b15f0dbd5 100644 --- a/block/bfq-cgroup.c +++ b/block/bfq-cgroup.c @@ -316,14 +316,12 @@ struct bfq_group *bfqq_group(struct bfq_queue *bfqq) static void bfqg_get(struct bfq_group *bfqg) { - bfqg->ref++; + refcount_inc(&bfqg->ref); } static void bfqg_put(struct bfq_group *bfqg) { - bfqg->ref--; - - if (bfqg->ref == 0) + if (refcount_dec_and_test(&bfqg->ref)) kfree(bfqg); } @@ -530,7 +528,7 @@ static struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, struct request_queue *q, } /* see comments in bfq_bic_update_cgroup for why refcounting */ - bfqg_get(bfqg); + refcount_set(&bfqg->ref, 1); return &bfqg->pd; } diff --git a/block/bfq-iosched.h b/block/bfq-iosched.h index 41aa151ccc22..466e4865ace6 100644 --- a/block/bfq-iosched.h +++ b/block/bfq-iosched.h @@ -928,7 +928,7 @@ struct bfq_group { char blkg_path[128]; /* reference counter (see comments in bfq_bic_update_cgroup) */ - int ref; + refcount_t ref; /* Is bfq_group still online? */ bool online;